AI Paper Atlas

Research map / Architectures and efficiency / RNNs and normalization

Symbolic Learning to Optimize: Towards Interpretability and Scalability

Wenqing Zheng, Tianlong Chen, Ting-Kuei Hu, Zhangyang Wang

ICLR 2022 · Architectures and efficiency › RNNs and normalization · approach: Neural-symbolic reasoning · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
Distill and meta-tune the learned optimizer as a symbolic ruleResNet-50 on CIFAR-10 · ResNet-50 on CIFAR-10test accuracy: 95.31 -> 95.40Improved
Use the distilled symbolic rule instead of AdamWBERT base cased · GLUE: SST-2, QQP, QNLI, MRPC, MNLI, CoLAtask metrics: SST-2 accuracy 89.15 -> 91.53; QQP F1/Accuracy 86.3/89.2 -> 87.8/90.8; QNLI accuracy 89.2 -> 90.1; MRPC F1/Accuracy 83.5Improved
Distill RP(small)(extra) into a symbolic equationRP(small)(extra) · P2-trained, P3 evaluationevaluation loss: 0.08 ± 0.05 -> 0.09 ± 0.04No clear effect
Distill DM into a symbolic equationDM · P2-trained, P3 evaluationevaluation loss: 0.23 ± 0.13 -> 0.33 ± 0.26Hurt
Use equation complexity 100 rather than 10 (ablation)Distilled RP(small)(extra) · ResNet-50 on CIFAR-10test accuracy: 93.5 -> 95.4Improved
Increase equation complexity from 100 to 200 (ablation)Distilled RP(small)(extra) · ResNet-50 on CIFAR-10test accuracy: 95.4 -> 12.0Hurt
Remove hyperbolic functions from the symbolic expression (ablation)DM distilled rule · P2R² score: 0.54 -> 0.53No clear effect
Meta-fine-tune the symbolic rule distilled from RP(small)(extra) (ablation)ResNet-50 on CIFAR-10, pretraining from P1/P2/P4 · ResNet-50 on CIFAR-10test accuracy: 91.3 -> 95.4; 93.8 -> 95.3; 94.4 -> 95.5Improved
Meta-fine-tune the symbolic rule distilled from DM (ablation)ResNet-50 on CIFAR-10, pretraining from P1/P2/P4 · ResNet-50 on CIFAR-10test accuracy: 13.4 -> 10.0; 70.0 -> 70.7; 83.2 -> 84.5Mixed
Add more diverse gradient-related input features (ablation)RP(small)(extra) versus simpler DM/RP variants · P2-trained, P3 evaluationevaluation loss: DM 0.23 ± 0.13; RP 0.12 ± 0.07; RP(small) 0.09 ± 0.05; RP(small)(extra) 0.08 ± 0.05Improved
Change the meta-pretraining problem while retaining the RP(small)(extra) feature design (ablation)RP(small)(extra) symbolic rule · ResNet-50 on CIFAR-10test accuracy after meta-fine-tuning: 95.4; 95.3; 95.5No clear effect

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy